关于「菜单甄别:Mussa–Rosen 三档与基础档不劣化约束」,《鲸鱼客群与尊享档》给出的定量结论是什么?
结论:第四章用穆萨与罗森的菜单甄别生成约 9.57 倍与 51.66 倍的档位价差,并指出不受约束的菜单让 0.575 的买家比现状更差;锁住基础档后利润仍保住 0.984,没有买家变差。
数据口径:核对原始文献之后做了蓝图修正:幂律必须先检验,一本纯对数正态的合成账簿通过了幂律拟合优度检验,p 值 0.430,只有似然比检验给出 z 为 -1.271、无法裁决;前 2 % 的利润份额随尾部形状从 0.121 变到 0.768,只能按区间报;「自动显影」用 BG/NBD 模型认出下一年前 2 % 客户的命中率是 0.625,不是 1;诱饵效应在自然呈现下可能减弱或反转;「隐式探针」改写为对所有访客公开、可自选的入口。 书中参数为示意,全部计算由书中程序生成。
对照数据表
下表数字由本章 Python 程序原样打印,可重新运行复现。 程序只用 Python 标准库与结论吻合的打印值: 0.575, 0.984, 51.66
| design profit base q, p plus p VIP p VIP/base | |
| status quo | 0.5000 1.000, 1.000 |
| U unconstr. | 0.8070 0.834, 0.834 7.983 43.083 51.66 |
| C locked | 0.7941 1.000, 1.000 7.552 42.652 42.65 |
| C tiers start at types t | 3.6 and t = 7.5 |
| U: share of buyers worse off than status quo 0.575 | |
| C: share of buyers worse off than status quo 0.000 | |
| C keeps 0.984 of U's profit, adds 0.2941 over status quo |
查看完整验算程序
# Chapter 4. p43: the no-degradation constraint. Adding plus and VIP
# tiers while the base tier keeps its old quality and price.
# Model: types and shares of p42. Status quo: one product; the best is
# q = 1, p = 1 sold to every type. Design U: the unconstrained best
# three-tier menu of p42 (base quality is cut). Design C: base locked at
# q = 1, p = 1; the plus and VIP tiers solve
# q_g = (t_g * S_g - t_(g+1) * S_(g+1)) / M_g, clamped from below, where
# t_g is the lowest type of tier g, M_g its share and S_g the share of
# tier g and above. A buyer is worse off if his best surplus falls.
import math
T = [1.0, 1.2, 1.5, 1.9, 2.6, 3.6, 5.2, 7.5]
F = [0.40, 0.25, 0.15, 0.09, 0.05, 0.035, 0.015, 0.01]
def tiers_profit(base, cuts):
qb, pb = base
bounds = list(cuts) + [len(T)]
groups = [list(range(bounds[i], bounds[i + 1]))
for i in range(len(bounds) - 1)]
menu = [(qb, pb)]
mass = [math.fsum(F[:cuts[0]])]
prev_q, prev_p = qb, pb
for gi, g in enumerate(groups):
m = math.fsum(F[i] for i in g)
s = math.fsum(F[g[0]:])
if gi + 1 < len(groups):
nxt = groups[gi + 1][0]
q = (T[g[0]] * s - T[nxt] * math.fsum(F[nxt:])) / m
else:
q = T[g[0]]
q = max(q, prev_q)
p = prev_p + T[g[0]] * (q - prev_q)
menu.append((q, p))
mass.append(m)
prev_q, prev_p = q, p
prof = math.fsum(m * (p - q * q / 2) for m, (q, p) in zip(mass, menu))
return prof, menu
def surplus(t, menu):
return max([0.0] + [t * q - p for q, p in menu])
if __name__ == "__main__":
quo = [(1.0, 1.0)]
quo_prof = math.fsum(F) * (1.0 - 0.5)
# design U copied from the output of p42
u_menu = [(0.834, 0.834), (2.82, 7.983), (7.5, 43.083)]
u_prof = 0.8070
best = None
for c1 in range(1, len(T) - 1):
for c2 in range(c1 + 1, len(T)):
pr, mn = tiers_profit((1.0, 1.0), (c1, c2))
if best is None or pr > best[0]:
best = (pr, mn, c1, c2)
c_prof, c_menu, c1, c2 = best
print("design profit base q, p plus p VIP p VIP/base")
for nm, pr, mn in (("status quo", quo_prof, quo),
("U unconstr.", u_prof, u_menu),
("C locked", c_prof, c_menu)):
extra = "" if len(mn) == 1 else "{:>7.3f} {:>7.3f} {:>6.2f}".format(
mn[1][1], mn[2][1], mn[2][1] / mn[0][1])
print("{:<12} {:>6.4f} {:.3f}, {:.3f} {}".format(
nm, pr, mn[0][0], mn[0][1], extra).rstrip())
print("C tiers start at types t = {:.1f} and t = {:.1f}".format(
T[c1], T[c2]))
for nm, mn in (("U", u_menu), ("C", c_menu)):
worse = math.fsum(f for t, f in zip(T, F)
if surplus(t, mn) < surplus(t, quo) - 1e-9)
print("{}: share of buyers worse off than status quo {:.3f}".format(
nm, worse))
print("C keeps {:.3f} of U's profit, adds {:.4f} over status quo".format(
c_prof / u_prof, c_prof - quo_prof))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《帕累托鲸鱼客群自动显影、高净值信号自筛选与十倍客单尊享层跃迁母典》第 4 章「菜单甄别:Mussa–Rosen 三档与基础档不劣化约束」(版本 1.0.0). 5D Library. DOI 待发布
DOI 待发布 · 专著页正式发布后上线(本站暂无专著页) · 5DL-B-399 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026399,
author = {Fan, Yafu},
title = {Pareto Whale Auto-Detection, High-Net-Worth Self-Selection Signals and Ten-X VIP Tier Escalation},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}